Movies can trigger perceptual, cognitive and emotional processes at multiple levels in the viewer's brain and thus provide useful tools to study human brain function. Furthermore, movies - as relatively natural stimuli - can help understand the brain activity supporting our everyday life. Earlier, brain activity during movie viewing has been studied mostly with functional magnetic resonance imaging (fMRI). Although fMRI provides a spatial resolution of millimeters, the temporal resolution of fMRI is limited, as fMRI measures the relatively slow changes of blood flow in the brain. Magnetoencephalography (MEG) measures the electromagnetic activity of synchronized neuronal populations with a temporal resolution of a millisecond. This Thesis focuses on finding consistent MEG signals across different movie viewers. In naturalistic experimental settings, it is challenging to uncover the signals of interest from the measured brain activity due to the com-plexity of both the brain function and the measured signals. Movies are also typically shown to the subjects only once, so that signal averaging cannot be used to enhance the signal-to-noise ratio. This Thesis aimed at developing MEG analysis tools that can capture complex brain responses to movies. Multiset canonical correlation analysis (MCCA) with spatial filtering, used to find consistent time-varying patterns of brain activity across viewers, played a major role in this Thesis. The first study addressed the similarity of MEG signals of eight movie viewers. Because the single-trial MEG signals are noisy, the previously used intersubject correlation methods did not uncover consistent brain activity across the viewers. However, MCCA-based spatial filtering allowed extract-ing signals that were significantly correlated across the subjects in several frequency bands. These signals originated from brain areas relevant for movie viewing, mostly visual cortices. The second study unraveled the dynamics of MEG signals originating in the somatosensory cortex during movie viewing. Tactile stimuli, presented once per second, probed the functional state of the somatosensory cortex during movie viewing. The evoked single-trial responses to these stimuli covaried with the appearance of haptic events in the movie. Thus, the movie contents continuously modulated the viewer's brain activity in a fine-grained manner. The third study compared MEG and fMRI signals of the same subjects during movie viewing. The correlations both across and within the subjects were stronger for fMRI than MEG signals. General linear modeling between the MEG and fMRI data demonstrated similarity between the electromagnetic and hemodynamic brain activity in occipital, temporal and frontal cortices. These studies demonstrate the feasibility of the developed and validated analysis approaches to uncover consistency in complex MEG signals measured during movie viewing. The results show that a part of the neuromagnetic brain activity is synchronized across viewers with high temporal accuracy in multiple brain regions.
- , Supervisor
- , Advisor
- , Advisor
|Publication status||Published - 2018|
|MoE publication type||G5 Doctoral dissertation (article)|
- brain, magnetoencephalography (MEG), functional magnetic resonance imaging (fMRI), brain imaging, canonical correlation analysis, naturalistic stimulation, movie